Talhue Search Overview Features Book a demo

Features

Everything Talhue Search does today.

What each feature does, how it works, and what recruiters see. Features that are still being built are marked.

Ask the way you think

Plain English in, precise logic out.

Type the request as you would say it to a colleague. Talhue pulls out the job title, skills, degrees, certifications, location and industry, and keeps your "and" and "or" exactly as written. It shows you how it read the query before it searches, so there are no hidden Boolean mistakes.

Example

software engineer or full stack developer in San Jose, CA or Austin, TX
with python and pandas or scikit

title    software engineer OR full stack developer
skills   (python AND pandas) OR scikit
places   San Jose, CA · Austin, TX

"python and (pandas or scikit)" keeps your parentheses; "python, pandas and scikit" means all three.
"bachelor's or master's in computer science" becomes two complete degree requirements.
Search one client's candidates or all of them.
Location and industry are read from the query today; filtering on them is coming next

Matching

AI reads every resume

One clear question per requirement.

Each requirement becomes a single yes-or-no question about each resume: "has the skill pandas?", "has the job title data scientist, or similar?". TypeSafe System One's Jev model answers from the whole resume, with a confidence. Talhue then applies your and/or logic in its own code and ranks the candidates, so the logic is exact and repeatable.

What every result shows

✓ exact match~ similar match✗ missing

You see why each candidate ranks where they do, requirement by requirement.

Similar titles count. Titles are always matched "or similar", so "Senior Backend Developer" matches "software engineer".
Include similar skills when you want reach: Angular for React, Pulumi for Terraform. Similar matches rank below exact ones.
Degrees match at the same level or higher in the same field; certifications match exactly.
Names, emails and phone numbers are removed before a resume is sent for scoring.
One question per requirement: single facts come back as a clear yes or no. One long compound question gives partial credit to near misses.

TypeSafe System One

Answers, not essays

Fast, low cost, no hallucinations.

Most AI search tools ask a large language model to read each resume and write about it. That is slow, expensive at the scale of a candidate database, and the written output can include things the resume never said. Talhue asks TypeSafe System One's Jev model structured questions instead.

Question type	Example	Jev returns
Yes or no	Has the certification CCRN?	Yes or no, with a confidence
Choice	Which role family does "Charge Nurse, ICU" belong to?	One of the listed families, with a confidence for each
Choice of numbers	How many years of professional experience?	A number of years, with how sure it is

Fast

A short answer comes back quickly, so every requirement is checked on every resume.

Low cost

You pay for reading resumes, not for writing text. Each search shows its cost.

No hallucinations

Answers can only be yes, no, or one of the options given. The confidence tells you how clear the resume was.

Talhue uses a general AI provider (OpenAI, Anthropic or Google) only where text has to be structured: turning your search into requirements, and listing the jobs on a resume.

Recency

Current experience, not old history

Is the candidate doing this work now?

For the top matches, Talhue checks whether each matched title and skill appears in the candidate's current role, their previous role, or only in older roles. The result is one label you can sort and filter by. For "or" requirements, the most recent option counts.

Label	Meaning
High	Every requirement is in the current role.
Medium	All requirements are in the current or previous role.
Low	Some requirements appear only in older roles.
None	The match comes only from older roles.

Hover a candidate to see their current and previous job, with company and years.

AI experience

Who actually works with AI

One ladder, every profession.

Clients increasingly ask for people who use AI in their work. Talhue scores every resume on the same ladder, with examples from many fields, so it recognizes AI experience in a nurse's or a marketer's resume as well as an engineer's. Filter search results by a minimum level.

0NoneNo AI experience on the resume.

1Uses AI toolsUses AI assistants for drafting, summarizing or coding.

2Applies AIPut AI into a team or customer process: AI lead scoring, AI-generated tests, an AI scribe in patient care.

3Creates AI solutionsBuilt, shipped or owned an AI or machine-learning product or feature.

4Leads AI adoptionLed AI adoption, training or strategy for a team or organization.

A measurable result from AI (time saved, revenue, accuracy) adds to the score.

Profiles

Every resume, read once

Structured data from free-form resumes.

When a candidate arrives from your ATS, Talhue reads the resume once and stores what it learns. Searches then use it without reading the resume again.

What	How	In search
Job history	Title, company and dates for every job, even with messy formats like "Spring 2019 – now".	Current and previous job shown on results
Years of experience	Total professional experience, counting overlapping jobs once and skipping gaps.	Filter coming next
Role family	Each job placed in a role family, such as Nursing or DevOps, Cloud & IT.	Filter coming next
Industry	Each employer given its official industry code (NAICS), for example hospitals or software publishers.	Filter coming next
AI experience	The level on the ladder above.	Filter and badge

Role families

Any role, and yours to extend

Start with 19 families. Add more in one file.

Talhue is not tied to tech hiring. Search works for any profession because the AI reads the resume, not a keyword list. Role families group jobs so you can report and filter by them.

Software EngineeringData Science, ML & AnalyticsDevOps, Cloud & ITQA & Test EngineeringHardware & EmbeddedData CenterProduct ManagementProgram & Project ManagementSalesMarketingCustomer Success & SupportFinance & AccountingHR & RecruitingOperations & Supply ChainNursingPhysicians: Primary Care & HospitalPhysicians: Medical SpecialtiesPhysicians: SurgicalOther

Add a role family for your market

Role families live in one plain-text file. Add a family with a few example titles, raise the file's version, and Talhue re-classifies the job titles it has seen. Where a title alone isn't enough, add a role-specific check. Talhue does this for data centers today: an "Electrician" counts as data center only if the resume shows data center work.

- id: pharmacy
  name: Pharmacy
  group: Healthcare
  example_titles: [Pharmacist, Clinical Pharmacist, Pharmacy Technician]

Setup

Up and running quickly

No training, no keyword lists.

There is nothing to teach Talhue before it works. It reads resumes as they are, in any profession, and recruiters search in their own words.

What a setup involves

Accounts: TypeSafe System One for matching, and one AI provider (OpenAI, Anthropic or Google).
Start the services as containers on your cloud or servers. Settings live in one file per service.
Connect your ATS. The connector sends candidates to Talhue and keeps them in sync.
Optional: add role families for your market, and choose an AI provider per client.

Primary Colors Technology can do the setup for you, including the ATS connector.

Your ATS

Fits into what you run

A simple API between your ATS and Talhue.

A connector reads candidates from your ATS on a schedule and sends them to Talhue. New candidates are added, changed resumes are read again, and unchanged ones cost nothing. We build the connector for your ATS.

Separate clients stay separate. Each client's candidates are searched and processed on their own.
Choose the AI provider per client: OpenAI, Anthropic or Google. A client's resumes and searches go only to its provider.
Runs where you want it. Containers for any cloud (AWS, Google Cloud, Azure) or your own servers.
Open source under Apache 2.0. Read the code, change it, and keep your changes.

Try it on your own candidates.

You need ATS API access, a TypeSafe account and one AI provider account. We do the setup, the ATS connector and the role families for your market.

Book a demo →