folk

What it is

folk is a per-seat CRM for teams that decided the CRM should be a shared contact database with a pipeline attached, not a platform with a full-time admin. The records are people, companies, and groups. The editing surface reads like a spreadsheet, and a new workspace is usable in an afternoon instead of a quarter. That is the whole argument, and it is the one folk makes in every thread about ripping out a heavyweight CRM: less product, less setup, less money.

The AI layer is not bolted on afterward. Magic Fields are prompt-backed columns, four named AI Assistants cover follow-up timing, interaction recaps, company research, and trigger-based email, and the folkX browser extension pulls contacts out of LinkedIn and seven other platforms. folk also runs its own hosted MCP server, which is still rare for a CRM at this price.

Why it shows up in RevOps stacks

• The MCP server is first-party and hosted. The endpoint is https://mcp.folk.app/mcp, authenticated over OAuth with no API key to manage, and folk’s docs list Claude Desktop, Claude Code, Cursor, VS Code, and ChatGPT as clients. The agent gets the same workspace access your own account has: search across people, companies, and groups; create and update contacts, notes, and custom fields; move deals through pipelines; and write relationship or pipeline summaries back.
• Magic Fields put a prompt in a column. You write one prompt that references the record’s own fields, and folk fills a value for every row — normalizing job titles, drafting a first line per contact, or classifying an account. The allowance is 2,000 magic fields per month on Standard and 5,000 on Premium.
• Capture happens where the contacts already are. folkX adds people from LinkedIn and seven-plus other sites in a click, and email, calendar, and WhatsApp sync into the contact record without a rep logging anything.

Pricing

Per member per month, with annual billing 20% under monthly. There is no free plan — a two-week trial runs on Premium features and asks for no card.

• Standard — $30 monthly, $24 annual ($288/member/year). Unlimited contacts, 2,000 messages per member/month, 500 enrichment credits per month workspace-wide, 2,000 magic fields/month. Automation reaches other systems through Zapier and Make only.
• Premium — $60 monthly, $48 annual ($576/member/year). Adds Deals, custom objects, email sequences, dashboards, REST API, advanced roles and permissions, and 3 custom domains. Raises limits to 5,000 messages per member/month, 1,000 enrichment credits, 5,000 magic fields.
• Enterprise — $100 monthly, $80 annual ($960/member/year). Custom limits, group-level permissions, security controls, custom integrations, dedicated account manager.

The real-world band is narrow because seats are the only variable that moves. Five people on Premium annual is $2,880/year. Ten people on Standard annual is also $2,880/year. The same ten seats on Attio Pro annual run $9,480/year before credit packs. folk’s published plans state monthly allowances with no overage rate attached, so budget the seat count and treat the allowances as ceilings rather than as a meter that will surprise you at renewal.

Best for

Founder-led and small GTM teams of 2–15 at pre-seed through Series A running relationship-driven sales, plus RevOps operators who want an agent reading and writing pipeline over MCP without standing up a middleware hop first.

Not for

Teams whose forecast is the deliverable. folk has pipeline and activity dashboards but no probability weighting by stage, no weighted pipeline view, no forecasting model, and no custom report builder. It is also wrong for teams that need automation to fire inside the CRM: moving a deal to a new stage triggers nothing natively, so that logic has to live in Zapier, Make, or n8n.

Versus the alternatives

• Salesforce — pick it when the evaluation turns on approval chains, territory governance, or a data model finance and support both read from. folk wins the first 90 days decisively; Salesforce wins the question of what happens at year three with 200 users.
• HubSpot — the free CRM tier plus marketing automation on one bill is the reason most teams land there. Pick HubSpot when marketing and sales must share a system; pick folk when you want a $24 record layer and explicitly nothing else attached to it.
• Attio — the other lightweight pole, at $35/seat annual on Plus and $79 on Pro, with custom objects behind Pro. Pick Attio when you are modeling a non-standard object graph or want credit-metered agent workflows inside the CRM. Pick folk when your objects really are people and companies and you want seats to be the only line item.
• Day.ai — the fastest-growing entrant in the segment, priced per agent at $24/$60/$200 with non-agent humans free. Pick Day.ai when agent count is the number you want on the invoice; pick folk when 10 or more humans need records, groups, and permissions of their own.

Watch-outs

• Enrichment credits are workspace-wide, not per seat. Standard’s 500 monthly credits cover the entire workspace, so ten seats share 500 and each rep effectively gets 50. Message allowances scale with headcount; enrichment does not. Guard: price Premium’s 1,000 credits against an external enrichment step in Clay before assuming folk covers the volume, and check the credit burn after one full month rather than at renewal.
• The advertised connector count is Zapier and Make reach, not native integrations. folk markets thousands of connected tools, but Standard’s programmatic access is Zapier and Make only and the REST API starts at Premium. Guard: list every system that has to talk to folk, mark each one native, Zapier-hop, or REST — if anything lands on REST, you are buying Premium at $48 annual, not Standard at $24.
• Deals are a Premium object, so part of the MCP surface has nothing to act on at the entry tier. The MCP server advertises moving deals through pipelines and keeping statuses in sync, and Deals do not exist on Standard. Guard: if the agent use case is pipeline hygiene, buy Premium from the start — MCP does not backfill an object the plan excludes.
• No forecasting and no stage-triggered automation. Both gaps are documented across review sites and neither closes with an upgrade. Guard: keep the forecast in a spreadsheet or BI tool, wire stage changes to n8n or Make, and confirm both paths work before migrating off a CRM that had them built in.
• Bulk import is the recurring complaint. Reviews across G2 and Capterra describe duplicate records after large imports and a tagging model that resists restructuring later. Guard: import one 200-row segment first, run folk’s dedupe on it, and settle the group and tag structure before loading the full contact book.