The people systems failure at most scaling startups is quiet and cumulative. In the first twenty employees, none of it matters: hiring is informal, compensation is negotiated case by case, onboarding is the founder walking a new hire through Notion for two hours, and performance management is a quarterly conversation over coffee. Nobody has written any of it down because there is nothing to write: each situation is handled as it comes up by people who know each other well. This works until it does not. At forty people the comp system breaks: two engineers at the same level learn they are paid differently and one of them leaves. At sixty people the org structure breaks: three teams are doing overlapping work nobody authorized and a fourth team is under-resourced because its headcount plan was never connected to the product roadmap. At eighty people the hiring breaks: the pipeline has six different interview processes depending on which hiring manager set it up, none of them have a scorecard, and the team is making inconsistent decisions that will not survive a legal challenge. By the time a company is large enough to hire a VP of People, the damage from five years of informal people systems is deeply embedded: pay inequity, role confusion, a culture that was never written down and is now being interpreted differently by people who were not there at the beginning.
The problem is not that founders are bad at people operations. The problem is that people systems require infrastructure the same way software requires infrastructure: stage-appropriate org design, explicit compensation bands, structured hiring pipelines with documented criteria, onboarding programs that get new hires productive in weeks rather than months, performance frameworks that give managers and reports a shared language for growth and expectations, and culture documentation that makes the implicit explicit before it becomes contested. Most early-stage companies have none of this because building it has always required a dedicated HR or People Ops function. That was the constraint. It is no longer the constraint.
Why people systems break at scale
The first failure is compensation. Most startups set early salaries by feel, by market data of varying quality, and by what candidates asked for. The result is a comp structure that is not a structure at all: a collection of individually negotiated outcomes with no relationship to each other and no principled basis for explaining why one person earns more than another at the same level. When the team is small and everyone knows how much the company is struggling, this tension stays latent. As the team grows and people start comparing notes, the latency ends. The damage is not just the departures: it is the trust that goes with them. Employees who discover pay inequity without a principled explanation do not just leave; they leave and talk. Compensation framework design is not an HR compliance exercise. It is a trust infrastructure project.
The second failure is org structure. Early teams organize around the people available, not around the work that needs to be done. A founding engineer becomes Head of Platform by default; a first marketing hire becomes Head of Growth without a decision about what growth actually means at this company; a senior product manager absorbs both strategy and execution because there is nobody else. These arrangements work when the team is ten people. By forty, the accidental org chart has become a source of political friction: reporting lines that do not match decision rights, spans of control that are too wide for the work to be done well, and team boundaries that create duplication and gaps simultaneously. Re-organizing a team that has grown into an accidental structure is significantly harder than designing the structure deliberately as the team scales.
The third failure is the one that is about to get much larger: the question of which roles in the company should still be filled by humans at all. Every operations function in a scaling company contains a mix of work that requires human judgment, relationships, and creativity, and work that is fundamentally information processing, status tracking, and document production. Until recently, the only way to get the second category done was to hire humans who would do the first and second categories together. That is no longer true. Human-to-agent migration is now a real planning discipline: auditing roles for automation potential, designing transition playbooks for the work that moves to agents, and building the hybrid teams that combine human judgment with agent execution at a fraction of the cost of hiring for pure headcount.
What a People Ops engineer actually does
A senior People Ops engineer is a systems architect for the human (and increasingly hybrid) layer of the organization. They design the structures, processes, and documentation that make a scaling company legible to the people inside it. They audit what exists, identify where the gaps are creating friction or risk, design replacements that are appropriate to the company's stage, and build the playbooks that make the systems run without requiring the architect to be present for every decision. They also think about the boundary between human and automated work with increasing precision: which roles are fundamentally relational and judgment-intensive, which roles are primarily procedural and information-intensive, and how the organization should plan for a world where the second category can be operated by AI agents running continuously at near-zero marginal cost.
Meet Folk
Folk is Tonone's dedicated AI People Ops Engineer: a purpose-built agent for the full people operations workflow at scaling companies. It does not produce generic HR policy templates or answer questions about employment law. It does the actual People Ops engineering work: auditing org design and hiring health, designing or reviewing org structure and headcount plans, building structured hiring pipelines with scorecards, designing compensation frameworks with salary bands and equity guidelines, creating employee onboarding playbooks, building performance management systems and career ladders, planning human-to-agent migration for roles with high automation potential, and documenting company culture in a form the organization can actually use. Folk gives scaling teams the people infrastructure that previously required a VP of People to build.
Tonone's Folk is the AI People Ops Engineer that audits your org structure, designs hiring pipelines with interview scorecards, builds comp bands by level and function, creates onboarding playbooks, designs performance frameworks, and plans which roles to migrate to AI agents so your people systems scale without a dedicated HR department.
What Folk actually does
Auditing people systems before building new ones
Before redesigning any people system, you need an honest picture of what your current systems are producing and where they are failing. The folk-recon skill audits the full people operations stack: org design health, hiring pipeline quality, compensation framework coherence, onboarding effectiveness, and performance management maturity. The output is a people ops health brief that names the specific failure patterns by domain. Which org units have spans of control that are too wide for the work. Where the hiring pipeline has inconsistent evaluation criteria across teams. Whether comp bands exist and whether they are applied consistently. How long it takes new hires to reach full productivity. Whether managers have a shared framework for performance conversations or whether each manager runs performance their own way. folk-recon prevents the team from building on a broken foundation: redesigning comp bands without addressing the role leveling problem that makes the bands ambiguous produces cleaner-looking spreadsheets with the same underlying inequity problem.
Designing org structure that matches the actual work
Org design is not a boxes-and-lines exercise. It is a decision about how information flows, where decisions get made, which teams own which outcomes, and how reporting structures create or prevent the collaboration the work requires. The folk-org skill designs or reviews org structure with explicit attention to spans of control, reporting lines, headcount plan, and team topology. The output is not a revised org chart; it is an org design specification: the rationale for each reporting relationship, the optimal span of control for each manager given their role's complexity and the team's current stage, the headcount plan that connects team growth to the product roadmap, and the team topology recommendation (whether work should be organized as stream-aligned teams, platform teams, enabling teams, or complicated-subsystem teams given the company's architecture and growth trajectory). folk-org also identifies where the current structure is creating friction: duplicated work across teams, decision-making bottlenecks where too many approvals are required for routine choices, and role boundaries that are so ambiguous that two teams are solving the same problem independently.
Building a hiring pipeline that produces consistent decisions
Most early-stage hiring pipelines are informal, inconsistent, and undocumented. The folk-hire skill builds a complete hiring pipeline for a role or function: a job description written for the candidate you actually want to attract (not a list of credentials), a sourcing strategy that identifies where that candidate actually spends their attention, an interview scorecard with explicit criteria mapped to the specific competencies the role requires, and a structured evaluation process that produces comparable signals across interviewers. The scorecard is the most important output. Without a scorecard, each interviewer evaluates the candidate against their own implicit standard: one interviewer weights technical depth heavily, another weights communication clarity, a third is testing culture fit by criteria they cannot articulate. The result is a debrief where three people have three strong opinions based on three different evaluations, and the hiring decision is made by whoever argues most persuasively in the room rather than by the data. folk-hire produces a scorecard that specifies exactly what each interviewer is evaluating, how evidence maps to ratings, and how the panel reaches a consistent decision.
Designing compensation bands that hold under scrutiny
A compensation framework is not a spreadsheet of salaries. It is a system of principles, levels, bands, and guidelines that produces defensible, consistent, and equitable compensation decisions at scale. The folk-comp skill designs a compensation framework from first principles: salary bands by level and function calibrated to a defined market percentile target, equity guidelines that connect grant size to level and stage with a coherent total compensation philosophy, and the decision framework that governs how individuals are placed within bands (not just at bands) based on scope, performance, and time in role. The framework includes the leveling criteria that define what it means to be an L3 versus L4 engineer or a Senior versus Staff product manager, because bands without leveling criteria are not bands: they are salary ranges that managers apply according to their own judgment, which reproduces the same inconsistency the bands were supposed to solve. folk-comp also produces the compensation narrative: the explanation of the total comp philosophy that managers can use in offer conversations and comp review discussions.
Building an onboarding program that gets people productive fast
The cost of a bad onboarding program is not just a slower ramp to productivity. It is the permanent first impression the company makes on the employee: the signal about whether the company is organized or chaotic, whether it has thought about the new hire's experience or expects them to figure it out, whether their manager is prepared to invest in their success or just needs another warm body in a seat. The folk-onboard skill builds a complete employee onboarding playbook from Day 1 through Week 4: a Day 1 checklist covering systems access, equipment, introductions, and the first context the new hire needs to feel oriented; Week 1 milestones covering the first meaningful work, team introductions, and key context documents; Week 2 through 4 progression covering deepening context and first independent contributions; and explicit 30-60-90 success criteria that give both the new hire and their manager a shared definition of what a successful ramp looks like. The playbook is role-type specific: onboarding for an engineer covers different ground than onboarding for an account executive, and folk-onboard produces role-appropriate versions rather than a single generic document that applies adequately to nobody.
Designing performance management that actually develops people
Performance management fails in two opposite directions. In the first direction, it is too subjective: managers give vague qualitative feedback that feels like assessment but produces no actionable direction for the employee. In the second direction, it is too mechanical: scores are assigned on a rating scale, calibration forces a distribution, and the process produces numbers that feel precise but are not meaningful. Neither approach develops people or creates shared understanding between managers and reports about what excellent performance looks like at each level. The folk-perf skill designs a performance management system that avoids both failure modes: a review cycle structure appropriate to the company's stage, a calibration process that ensures consistency across managers without forcing artificial distributions, a career ladder that specifies what progression from one level to the next looks like in behavioral and outcome terms, and a goal-setting framework that connects individual work to team and company outcomes. The career ladder is the foundation: without explicit level definitions and progression criteria, performance conversations are inevitably about manager opinions rather than evidence mapped to a shared standard.
Planning which roles to migrate to AI agents
This is the skill that makes Folk genuinely different from every HR tool and generalist AI that came before it. The folk-migrate skill is a human-to-agent migration planner: it audits roles in the organization for automation potential, designs transition playbooks for work that can move to AI agents, and produces a migration roadmap that gives the leadership team a staged, realistic plan for moving from human-only execution to hybrid human-agent teams. The audit examines each role along two dimensions: the proportion of the role's work that is information processing, status tracking, documentation, and structured communication (high automation potential), and the proportion that is relationship management, judgment under ambiguity, and creative problem-solving (low automation potential). Roles with a high proportion of the first category are candidates for full migration to AI agents. Roles with a mix become hybrid roles where humans focus on the judgment-intensive work and agents handle the procedural volume. folk-migrate produces not just an assessment but a transition playbook for each migrating role: which Tonone agents replace which work, what the agent configuration looks like, what new human responsibilities emerge as a result of the migration, and what the cost and capacity model looks like at steady state. For a company planning to migrate 3 operations roles to AI agents, folk-migrate turns an ambiguous ambition into a specific, sequenced plan with clear decision points and success criteria at each phase.
The folk-migrate skill is most valuable when run before a headcount decision, not after. If you are debating whether to hire a third ops coordinator or a second customer success manager, run folk-migrate on those roles first. The output will tell you exactly which parts of the role are high automation potential, which Tonone agents can cover them today, and what the headcount decision actually is once you separate the automatable work from the judgment-intensive work that still needs a human.
Documenting culture before it becomes contested
Company culture is the set of behavioral norms, decision-making patterns, and values that determine how work actually gets done when nobody is watching. In the first twenty employees, culture is transmitted by proximity: new hires observe how founders make decisions, handle conflict, celebrate wins, and respond to failure. They absorb the norms directly. By sixty employees, that transmission mechanism breaks down: new hires cannot observe the founders in every context, and the behaviors they see from mid-tenure employees may already be drifting from the original norms. Culture does not break all at once; it erodes gradually, replaced by the locally optimal behaviors of whoever is in the room. The folk-culture skill documents and strengthens company culture: it codifies values in specific behavioral terms (not generic statements like 'we value transparency' but specific descriptions of what transparency looks like in practice at this company), documents the behavioral norms that distinguish this company's operating style, and produces a culture health check that identifies where actual observed behavior is diverging from stated values. The output is a culture document that is genuinely operational: managers can use it in hiring to evaluate cultural fit against specific criteria, new hires can use it to calibrate their own behavior, and leaders can use it in performance conversations to address norm violations with a shared reference point.
Worked example: scaling from 50 to 120 people
Consider a Series B startup with 50 people preparing to scale to 120 over the next 18 months. The fundraise closed six months ago; the engineering and sales teams are both hiring aggressively. The founding team has identified three specific problems: the hiring process is inconsistent and recently produced two mismatches in senior hires that cost the company six months of onboarding time and two expensive separations; compensation is a mess with no formal bands, and two senior engineers have already flagged the issue after comparing offers with candidates; and the leadership team believes three of the current ops roles could probably be handled by AI agents but has no framework for deciding which parts of which roles or how to sequence the transition. The company runs folk-recon first.
The folk-recon output identifies five specific issues: (1) The hiring pipeline has four different interview processes across engineering, sales, and operations, none with a written scorecard, and the debriefs are unstructured discussions rather than score-based evaluations. (2) Compensation is entirely negotiated: the range between lowest and highest paid engineers at the same experience level is 40%, and there is no documented rationale for any individual placement. (3) Onboarding is a Notion wiki that was last updated nine months ago and does not reflect the current product, team structure, or tooling. (4) Performance management does not exist in any formal sense: two managers do quarterly reviews, two do annual reviews, and two have never done a structured review. (5) The culture deck from the seed round is four slides of values statements with no behavioral specificity, and three managers have recently described the company's culture in ways that contradict each other. The team uses folk-recon as the brief for a five-week People Ops build sprint.
Week one: folk-hire produces a standardized hiring pipeline for each of the three major functions, engineering, sales, and operations. Each pipeline has a job description framework, a sourcing strategy, and a structured interview scorecard. The engineering scorecard has four interview stages with explicit competency maps: a technical screen scored on problem decomposition and code quality, a system design interview scored on trade-off reasoning and communication clarity, a behavioral panel scored on collaboration, conflict resolution, and learning orientation, and a values interview scored against the culture criteria from folk-culture. The debrief process produces a score per stage per competency with a documented hire, no-hire, or more-data decision.
Week two: folk-comp produces a compensation framework with salary bands at five levels for engineering and three levels for sales and operations, calibrated to the 75th percentile of the relevant market. The framework includes equity guidelines at each level for new grants and refresh grants, a leveling rubric that specifies the scope and impact expectations for each level, and a placement philosophy that determines where within a band an individual is placed based on scope match, performance signals, and time in role. The framework resolves the immediate pay equity issue: three engineers move to the midpoint of their correct band; two receive retroactive adjustments. The company now has a defensible answer when an engineer asks why they are paid what they are paid.
Week three: folk-migrate audits the three operations roles identified by leadership as migration candidates. Role one is a data operations coordinator whose work is 70% data entry, pipeline management, and report generation and 30% vendor relationship management and exception handling. The assessment: migrate the 70% to a combination of Tonone's Flux (data pipelines) and Lens (reporting) agents; retain the coordinator's headcount but redesign the role around vendor management, exception handling, and quality oversight of the automated pipelines. Role two is an internal IT and systems access coordinator whose work is 85% ticket processing, access provisioning, and software license management. The assessment: full migration to Tonone's Pave agent for environment and access management; the role is eliminated from the next hiring cycle and replaced with a quarterly systems audit by a contracted IT consultant. Role three is a customer onboarding coordinator whose work is 55% email coordination, 20% document production, and 25% relationship management and account escalation management. The assessment: migrate coordination and document production to Tonone's Keep agent; retain a human for relationship management and escalations. The migration roadmap sequences the transitions over 12 weeks with a specific agent configuration guide for each.
Week four: folk-onboard produces role-specific onboarding playbooks for engineering, sales, and operations. Each playbook covers Day 1 through Week 4 with daily milestone checkpoints, systems access timelines, first-work assignments, and explicit 30-60-90 success criteria. The 30-day criteria for an engineer: has shipped one feature to production, has attended five architectural decision discussions, can navigate the codebase without assistance, and has had the first 1:1 structured conversation with their manager about their 90-day goals. The 90-day criteria: is an independent contributor on a core product area, has shipped at least three production changes with minimal review, and has received a formal mid-ramp feedback session with ratings on the scorecard competencies used in hiring. folk-perf completes the sprint with a performance management system including a semi-annual review cycle, a calibration process, and an engineering career ladder from L1 through L6 with behavioral and outcome criteria at each level.
Run folk-org before you start the next hiring wave, not after. The most expensive org design mistakes are the ones baked into headcount plans: a manager hired to run a team that should not exist as a separate unit, a function that duplicates work already owned by another team, or a span of control so wide that the manager cannot actually develop their reports. folk-org reviews your current structure and proposed headcount plan and identifies these problems before the hires are made and the reporting lines are set.
Folk vs the alternatives
Folk is not an HRIS that stores employee records. It is not an ATS that tracks applications. It is not a generalist chatbot that answers HR policy questions. It is a People Ops engineering agent that designs the systems your organization runs on: org structure, hiring infrastructure, comp frameworks, onboarding programs, performance systems, culture documentation, and migration plans for transitioning human work to AI agents. The comparison below makes the functional differences specific.
| Capability | Tonone | Generalist chatbot | Cursor / Copilot |
|---|---|---|---|
| Org design with headcount planning | Yes, folk-org designs or reviews org structure with spans of control, reporting lines, headcount plan connected to roadmap, and team topology recommendation with rationale | Describes org design principles generically; does not produce a specific org structure recommendation with headcount plan and team topology for a given company context | BambooHR and Rippling store the org chart and headcount data; org design decisions require human analysis outside the platform |
| Hiring pipeline with interview scorecards | Yes, folk-hire builds job description, sourcing strategy, interview scorecard with competency maps, and structured evaluation process per role and function | Writes job descriptions and interview questions from prompts; does not produce a structured pipeline with competency-mapped scorecards and debrief process | Greenhouse and Lever track candidates through a defined pipeline; scorecard design and competency mapping require manual configuration by the People team |
| Compensation framework with salary bands | Yes, folk-comp designs salary bands by level and function, equity guidelines, total comp philosophy, leveling rubric, and placement framework | Discusses comp philosophy and can suggest band ranges; does not produce a full compensation framework with leveling criteria and placement guidelines | Radford, Levels.fyi, and Carta provide market data; comp framework design (bands, leveling, philosophy) requires separate People Ops work to build |
| Human-to-agent migration planning | Yes, folk-migrate audits roles for automation potential, designs transition playbooks, identifies which Tonone agents replace which work, and produces a sequenced migration roadmap | Can discuss automation potential conceptually; does not produce a role-level audit with transition playbooks and agent configuration guides | No HRIS or HR tool includes human-to-agent migration planning; this capability does not exist in any existing HR platform |
| Employee onboarding playbooks | Yes, folk-onboard builds Day 1 through Week 4 playbooks by role type with daily milestones, systems access timelines, and explicit 30-60-90 success criteria | Writes onboarding checklists from prompts; does not produce role-specific playbooks with structured milestones and calibrated success criteria | BambooHR Onboarding and Rippling automate task assignment and systems provisioning; playbook content design and success criteria require separate People Ops work |
| Performance management with career ladders | Yes, folk-perf designs review cycles, calibration process, career ladder with behavioral and outcome criteria per level, and goal-setting framework | Writes performance review templates and discusses career ladder principles; does not produce a full performance system with calibration process and level-specific criteria | Lattice and Leapsome run the review cycle and collect feedback; career ladder design and calibration framework require People Ops work to build before the tools can use them |
Tonone's folk-migrate skill is the first AI capability that audits roles for automation potential, designs transition playbooks for work that moves to AI agents, and produces a sequenced migration roadmap with agent configuration guides so scaling teams can plan human-to-agent transitions with the same rigor they apply to product roadmaps.
Why folk-migrate is the most important skill in the suite
Every other skill in Folk builds infrastructure for a team of humans. folk-migrate builds infrastructure for the transition from a team of humans to a hybrid team of humans and agents. This is not a marginal capability; it is the central question facing every operations function in the next three years. The companies that answer it well will have significant structural cost advantages over those that do not: not because they are cutting headcount indiscriminately, but because they are focusing human work on the parts of each role that genuinely require human judgment, and routing the procedural, information-intensive work to agents that run continuously at near-zero marginal cost.
The folk-migrate approach is role-level and function-level, not task-level. The question is not 'can an AI write this email' but 'what is the actual composition of this role, what proportion of it is high automation potential, which Tonone agents replace which specific work streams, and what does the redesigned human role look like when the automatable work is removed.' This is a planning discipline, not a technology question. Most companies are not asking it rigorously because they do not have a framework for the analysis. folk-migrate provides the framework, applies it to specific roles, and produces actionable plans rather than philosophical positions about AI and the future of work.
The migration roadmap output is sequenced by three criteria: automation potential (how much of the role is high-confidence automatable today), organizational risk (how critical is this function to day-to-day operations and what is the blast radius if the migration creates problems), and agent readiness (which Tonone agents are mature enough to reliably replace the relevant work streams). The sequencing produces a 12-to-24 week plan that moves the highest-potential, lowest-risk migrations first, building organizational confidence in the approach before tackling more complex transitions. Each phase has explicit success criteria and a rollback plan if the agent performance does not meet the threshold. This is migration engineering, and it is what separates a thoughtful transition from an experiment that damages the operations function.
Install and try
Tonone is free and MIT-licensed. Install it once and all agents, including Folk, are available in your Claude Code session immediately. You pay only for Claude Code token usage during the work itself. Start with folk-recon to get an honest picture of your people systems before building anything new. If you are making a headcount decision right now, run folk-migrate on the role before posting the job description. If you have engineers asking about compensation or equity, run folk-comp to produce a defensible framework before the next conversation.
1. Add to marketplace
2. Install Folk
Frequently asked questions
- What does Tonone's Folk agent do?
- Folk is Tonone's AI People Ops Engineer. It audits org design, hiring pipeline health, and compensation frameworks; designs org structure and headcount plans; builds structured hiring pipelines with interview scorecards; designs compensation frameworks with salary bands and leveling criteria; creates role-specific onboarding playbooks with 30-60-90 success criteria; builds performance management systems and career ladders; plans human-to-agent migration for roles with high automation potential; and documents company culture in operational behavioral terms.
- What is the folk-migrate skill and how does it work?
- folk-migrate is Folk's human-to-agent migration planning skill. It audits roles in the organization for automation potential by examining what proportion of each role's work is information processing, status tracking, and document production (high automation potential) versus relationship management, judgment under ambiguity, and creative problem-solving (lower automation potential). For roles with high automation potential, it designs a transition playbook specifying which Tonone agents replace which work streams, what the redesigned human role looks like, and what the cost and capacity model is at steady state. The output is a sequenced migration roadmap with success criteria and rollback plans at each phase.
- How does Folk's compensation framework design work?
- The folk-comp skill designs a compensation framework from first principles: salary bands by level and function calibrated to a defined market percentile target, equity guidelines connecting grant size to level and company stage, a leveling rubric defining what each level means in scope and impact terms, and a placement framework governing where within a band individuals are placed based on scope match and performance evidence. The framework also produces the compensation narrative managers can use in offer and comp review conversations.
- How is Folk different from BambooHR or Rippling?
- BambooHR and Rippling store and automate people operations workflows: they track headcount, run review cycles, automate onboarding task assignment, and manage employee records. They do not design the org structure, compensation framework, career ladders, or hiring pipelines that those workflows operate on. Folk designs those systems. It is the planning and architecture layer; HRIS platforms are the execution and record-keeping layer.
- Can Folk help a 50-person startup that has never had formal HR?
- Yes. Folk is specifically designed for the stage between informal founder-led people management and a full HR department. Run folk-recon first to get a health brief identifying the highest-priority gaps. For most companies at this stage, the sequence is: folk-comp to address immediate pay equity risk, folk-hire to standardize the hiring process before the next hiring wave, folk-onboard to improve ramp time for people already joining, and folk-migrate to make the next headcount decision with a clear view of what should be a human role versus an agent configuration.