Umamy
Agentic hiring workspace

The best people aren’t looking.
They’re listening.

Agents map everyone who could do the job, read who is actually open to move, and write from your team’s seats. You approve every message.

Works with your ATS and LinkedIn seats~4 weeks to a signed hire
Senior Backend Engineer · Paris
Activity
11,686A-players foundscored 80+ on a real scorecard
1 in 2replied4,484 of 9,242 people messaged
1,056asked to talksaid yes in their own words, not a click
~4 weeksto a signed hirehalf the time a firm takes
How it works

Live in an afternoon, working by the morning

No migration, no new pipeline to learn. The agents plug into what you already use and start on the role you already have open.

01

Connect your tools

Your ATS, your LinkedIn seats, your note-taker. Read-only until you say otherwise, and you pick which tools the agents may call.

Connected
AshbyLinkedInNootaGmail
Read-only by default · 31 connectors
02

Define what great looks like

One call turns your criteria into a scorecard. Every candidate is then measured against it, not against a keyword.

Scorecard
Senior backend7+ years in GoSeries B or later
5 must-haves and 3 outcomes, agreed on the call
03

Agents work, you decide

Sourcing, screening and booking run every day. Drafts wait for your approval, and the whole trail stays on the record.

Agents runningday 2
Open to move
Watching
3 drafts waiting for you
Product

The surfaces your team lives in

See all 31 integrations
IntegrationsEvery tool an agent can use, and what it’s allowed to do.
Add a tool
LinkedIn seats · agents source from these profiles
Your founder
Sends from their own profile · 40 a day
Connected
Your talent lead
Writes and approves · 25 a day
Connected
Your hiring manager
Reviews the shortlist · 15 a day
Connected
Gmail
Send approved sequences, read replies
Connected
31
Google Calendar
Book interviews on real availability
Connected
Noota
Interview notes onto the candidate record
Connected
Ashby
Reads your pipeline, writes candidates back
Connected
Granola
Meeting notes from your hiring syncs
Connect
Company context
Scorecards, comp bands, interview loops
12 docs
Plus Slack, Lever, Teamtailor, Workable and MCP —browse the directory
What the agents actually do

The work a recruiter does, done on the record

Source and map talent pools

The whole market for a role, not a keyword list — with a mobility read on every relevant profile, so you see who is actually open before anyone is contacted.

1,482 mapped34 companies212 on-profile
Who is actually open
Moving now41
Open to move118
Worth watching213
Settled1,110
Top of the pool
Open
Watching
a mobility read on all 1,482

Rediscover people you already have

Agents re-read every past applicant in your ATS, find what changed since, and flag the ones who fit the role you opened today.

Past applicantApplied 2024
rejected then
What changed since
Promoted to Staff Engineer · payments
Comp signal now €80–90k, inside band
Fits the role you opened today
1,140already screened4 minto re-score them all

Pick up where the agent left off

Step into any conversation and everything the agents learned is already on the candidate. No thread to re-read before your call.

Agent · first message sent Mon
Agent · screening done Wed
Notice 2 months · €78k · remote 2 days
You take the call Thu
Brief already written

Run screening conversations

Notice period, comp, must-haves — and the technical questions you could not ask yourself. Hiring a data scientist without being one, the agent probes the models they shipped and how they validated them.

14hof screening calls you did not take, last month
Wherever they reply
Screening · data scientist
Which models did you take to production, and how did you validate them?
Shipped two ranking models, validated on held-out quarters. Recommended for shortlist.Fit 91
The mobility read

Who is open, not just who fits

Tenure, switch cadence, comp gap, relocation history. You decide what counts for this role — the read updates as you change your mind, and nobody is contacted on a guess.

Try it: turn a factor off and watch the read move.

Not contacted
Open to movemobility read
SettledWorth watchingOpen to moveMoving now
Your rules for this role
Tenure & switch cadence3y 11m in role · moves every ~3y
Comp band gap~12% under market for the level
Relocation & remoteParis → open, remote-first history
388 of 1,482 profiles read as open to move or better
Before you spend a call

Know who is actually interested

A reply is not a yes. Agents turn each one into the facts that decide your next move — available, open to relocate, wants a time — so you know who to call first and nobody is asked the same question twice.

“yes i'm up for a chat.”InterestedAvailable to chatPositive sentiment
“Yes, I'm open to relocate”InterestedOpen to relocationGeo-flexible
“tomorrow at 3:30 works fine”Booked meetingProposing timesMeeting booked
“Sounds interesting, could you send me the job spec?”Requested infoWants the detailPositive sentiment
3,444 signals labelled so far99.85% without a human reading themReal replies, shown truncated and without names
Why it gets better

Every hire makes the next one faster

An agency starts each search from zero. Every reply your agents get is labelled and kept, so the second role in a market is quicker than the first — and the third is quicker again.

10,074candidate replies capturedthe raw material nobody else has
3,444intent signals labelled99.85% of them by the agents
716people found availableworth a message right now

13,083 messages sent so far — the same output as two or three recruiting firms, turned into an asset that keeps working after the hire is closed.

Integrations

Hire from the AI you already talk to

Ask in Claude or ChatGPT, or let it come to you in Slack. The agents read your pool, check who is open, and draft — you approve.

ClaudeClaude
ChatGPTChatGPT
Slack
Gemini, Perplexity, Le Chat, Teams
Asking an assistant for three backend engineers open to move
And works inside the tools you already pay for
LinkedIn
Recruitee
Lever
Ashby
Teamtailor
Workable
Gmail
Noota
Granola
22 more
Control

Agents that ask permission, not forgiveness

Automation is only useful if you’d defend it to a candidate. Limits you set, actions on the record.

Clear approval points
Set gates per hunt: first send, follow-ups, scheduling, ATS writes.
Full visibility into agent actions
Every search, message and ATS write is logged with its reasoning. Exportable.
Humans make the calls
Agents qualify and organize. Interviews and offers stay with your team.
Audit log · Senior Backend EngineerRead-only
Ariel approved 3 interview invitesApproval gate cleared
09:29
Sourcing agent skipped 14 profilesYour do-not-contact list
09:58
ATS agent wrote 4 notes to AshbyScoped to this role only
09:41
+1
3 people can approve. Every action is attributed and exportable.
Your brand, their first impression

Candidates hear from your team, not an agency

Agents draft; the message goes out from your CEO, VP or hiring manager once you approve it. That’s why good people reply.

Candidates talk directly to decision makers
Full transparency on the role and the process
Every candidate gets an answer
Fewer steps, faster decisions
Messages drafted by the agents, waiting for your approval before they send
up to 95%connection acceptance
49%average reply rate
up to 69%from a founder or hiring-manager seat

Aggregate across 9,242 candidates messaged. The higher figures come from deployments where the founder seat is live and the scorecard is tight.

For hiring leaders

More capacity, no more headcount

Three roles at once, no new recruiter, no agency retainer.
Flat pricing, no success fees.
One view of what moved on every role.
~4 weeksto a signed hire
For founders and hiring leaders
For talent teams

Fewer admin calls, richer context

Skip the notice-period-and-comp calls.
Evidence, transcript and a fit score you can argue with.
Your ATS stays the source of truth.
1 in 2of people messaged reply
For talent teams
800M+profiles reachable
20+ATS integrations
11,686A-players found
99.85%labelled without a human

“We are currently scaling really fast and we need to recruit the best. Umamy enables to do it ourselves and to keep the control on our hirings!”

Edouard Bouvet
Edouard Bouvet
Co-Founder & CEO · Swyt Solutions

“We have various needs, and we have tried going through headhunters in the past. We were lacking quality and transparency. With Umamy we get it all back and more.”

Thomas Reynaud
Thomas Reynaud
CEO & Co-founder · Garantme

“I didn’t have time to review every profile and felt like I was missing out on great people. Umamy scores everyone fairly and surfaces the best candidates for me.”

Frédéric Clément
Frédéric Clément
Co-Founder & CEO · Mimbi
Mimbi

“We kept interviewing the wrong people and wasting hours on mismatched candidates. Umamy lets us control quality from the start and only meet the right profiles.”

Sylvain Demortier
Sylvain Demortier
CEO · LynxUp Consulting
LynxUp
GarantmeTimeleftSundaySwytiAdvizeNebestaMimbiKamet Ventures
Who builds the agents

Engineers who have hired. Recruiters who ship software.

We build the agents ourselves and we have filled the roles they work on. That is why they behave like recruiters instead of chatbots.

Backed byCegid Startup ProgramHEC Incubateur
Stéphane Millet
Stéphane MilletCEO

Two-time founder, exits in 2016 and 2021. Invested in 17 early-stage companies.

Clément Revillion
Clément RevillionCSO

Closed €10M+ enterprise deals at Google and Salesforce.

Connor Blair
Connor BlairCTO

ML engineer, UIUC-trained. Builds systems that ship fast and stay measurable.

Jonathan Bouaziz
Jonathan BouazizCPO

Engineer turned product. Ex-Pong.ai, 7 years in NYC, Apple Best App winner.

Give your agents something to do this week

Connect your ATS, write one hiring goal, approve the first messages.

Frequent questions

Straight answers

Anything we have not covered? Book a call and ask us directly.