Teamflect runs performance reviews inside Microsoft Teams

Teamflect runs performance reviews, goals and 1-on-1 meetings inside Microsoft Teams and Outlook. It is free for up to 10 users. Paid plans start at $7 per user per month, billed annually. It needs a Microsoft 365 account.

By Ashish Manchanda Published 16 Oct 2026 Last updated 16 Oct 2026

Teamflect suits companies that work in Microsoft Teams, if the AI plan and the contract fit

In one line

Teamflect runs review rounds, goals, feedback and 1-on-1 meetings inside Microsoft Teams. Its AI features draft review comments, summarise feedback and suggest places on a nine-box grid.

Buy it if

Your staff already work in Microsoft Teams. You want to try a full review system free with up to 10 people. You can sign an annual contract that does not let you cut licences mid-term.

Do not buy it if

Your company works first in Google Workspace or Slack. You need one product that also runs payroll, benefits and core HR records. You need unlimited use of the AI Agent without paying for Professional.

Check first

Ask Teamflect which AI features come with your plan. Its AI page and its help centre give different answers.

Price

Starter is free for up to 10 users. Essential costs $7 per user per month billed annually, or $9 billed monthly. Professional costs $11 billed annually, or $14 billed monthly. Enterprise is priced by quote.

How a review round runs inside Teams

Teamflect lives in Microsoft Teams and Outlook. Staff open it inside Teams. On a phone they use the Microsoft Teams app or a mobile browser. They sign in with their Microsoft 365 login. Teamflect uses Microsoft Graph permissions to connect to Teams, Outlook, Entra ID and To Do. Entra ID is Microsoft’s staff directory.

An administrator builds a review template and starts a round. The template sets the flow. A manager can review without a self-review. Or the employee writes a self-review first and the manager adds comments on top. Rounds can repeat weekly, monthly, quarterly or yearly. Probation reviews can launch a set number of days after hire. Rating questions use a 5-emoji scale by default, or a 3-emoji scale. A question can require a comment, for example when someone gives 1 out of 5. The overall rating comes from named criteria with weights. The goal-progress criterion is calculated automatically, and the reviewer cannot change it. A review can carry an electronic signature. It can also be hidden from the employee, who then gets no notice.

A 360-degree section collects feedback from people who work with the employee, with at most five feedback providers. Anonymous comments appear as a summary, not one by one. Calibration balances scores across teams to prevent rating bias. With calibration on, finished reviews go to administrators first. They can enter calibrated scores, and the employee sees nothing until calibration ends. A Calibration Supervisor or the reviewer can finalise it. Managers and admins can compare teams’ average ratings side by side. Results can also appear on a nine-box grid, a chart that places each person by performance and potential.

Goals set in the goals area can be pulled into a review as a question set. A 1-on-1 meeting has a custom agenda, shared notes and private notes. Illinois Heartland Library System is a customer. It reports that 1-on-1 conversations recorded in Teams are added to the review automatically. Admins decide whether managers can see their direct reports’ private data.

Professional plans add a link to an HR system. It reads the company directory, the organisation structure, and employment and income data. It connects one system at a time and pulls updates every 24 hours. Feeding changes back can take up to 48 hours. It connects to more than 200 HR systems, including Workday, UKG, ADP, BambooHR and HiBob. Professional also includes a connector to Microsoft’s Power BI reporting tool.

Global administrators can create API keys. An API is a way for your own systems to connect to Teamflect.

Professional lets other AI assistants act inside Teamflect. The link uses a standard called MCP, short for Model Context Protocol. It lets staff connect assistants such as Claude Desktop, Claude Code and ChatGPT to Teamflect. Its tools can fetch review rounds, completed review forms and nine-box placements. Its write tools let the connected assistant perform supported actions in Teamflect. Turning it on adds no extra Teamflect permissions for users. Full ChatGPT support is in beta, which means a test version. It covers tools that create or change information. It is for ChatGPT Business and Enterprise on the web. You can try the MCP link free for 30 days.

Other products in this market treat AI in reviews differently. You can compare how they handle it in our performance management software guide.

The AI drafts, admins hold the switches and you can start free

It sits where Microsoft 365 staff already work

Staff reach reviews, goals and feedback from Teams and Outlook. They sign in with their Microsoft 365 login. A CSV file (a spreadsheet file) or an Entra ID sync brings in goals, employee lists and review history. Essential comes with free help to set up and personal training sessions. A customer success manager helps set up templates, move over old processes and train managers.

The AI helps with the wording and the reviewer decides

Enhance with AI rewrites an answer to be more concise and professional. It has a Remove Biased Language option. The writer can compare the new version with the old one before using it. It works in self-reviews, manager reviews and peer reviews. Admins choose who can use it, the reviewee only, the reviewer only or both. The nine-box suggestion does not submit itself, override the reviewer or change employee data. The reviewer can pick another box. Teamflect advises treating AI feedback summaries as a starting point, not a final judgement.

Admins hold the AI switches

A master switch turns off every AI feature. Admins can also manage each feature on its own. They can turn the AI Agent on or off and choose who can use it. Teamflect processes customer data with AI only when the customer turns a feature on and submits data. The customer can ask Teamflect to turn off AI processing for specific people. Teamflect uses OpenAI models hosted on Microsoft Azure. They run in a regional data centre near the customer, depending on the AI provider’s capacity.

The AI Agent works in Teams and Copilot

Staff click Ask AI in the Teamflect app. The AI Agent is also in the Microsoft Teams chat and is a plug-in for Microsoft Copilot. It drafts self-review and manager comments from goals, tasks, 360 input and feedback in the review period. Both sides edit before they submit. It flags biased or legally sensitive language. It prepares 1-on-1 meetings from goals, feedback and survey results. It answers from a connected HR system or from handbooks that an admin uploads and switches on. It looks for patterns in feedback, goals, tasks and meetings to help managers spot burnout and hidden high performers.

Independent security checks

Teamflect holds SOC 1 and SOC 2 Type I and Type II certification. SOC reports are independent audits of security controls. Teamflect also complies with GDPR, the EU data protection law. Independent experts run biannual penetration tests, which are simulated attacks on its systems. Teamflect shares the results publicly in its Trust Center.

You can try all of it before you pay

Starter is free for up to 10 users. It includes all nine parts of Teamflect, which it calls “modules”. New users get a 30-day Professional trial and then drop to Starter.

Customers from 10 people to more than 1,000

Staff who already work in Microsoft Teams

Teamflect was built natively for Microsoft Teams and Outlook. Companies that work first in Google Workspace or Slack do not get the benefit of its native integration. Teamflect scales from a 10-user starter plan to rollouts across thousands of employees in several countries.

Customers of many sizes

Ultimate Technologies Group has 10 to 50 employees. Horizon House has about 250. Montego has over 800. Decospan has more than 1,000. Furman University is in the 1,001 to 5,000 band. Securitas ran a pilot within Securitas Services Europe. Securitas Europe has 120,000 employees across 20 countries. Invacare operates in 22 countries. Teamflect reports that a department within Furman University achieved 50% of its strategic goals within a year.

A small team that wants less manual HR work

Ultimate Technologies Group reports nearly 66 percent less time on manual HR work after adopting Teamflect. It estimated it would have needed at least three more HR hires without it. DPC Technologies used Teamflect after a merger that doubled its team from 20 to 40 employees.

A team that wants AI help in every review

NCDMM is a customer with 51 to 200 employees. Its IT director said nobody did a review without clicking Enhance with AI.

Teamflect’s About page puts the number of organisations at more than 1,000. The Microsoft Marketplace listing puts it at more than 5,000 companies. Teamflect’s AI Agent page quotes 4.6 out of 5 on Microsoft AppSource, and its Enterprise page quotes 4.7. The Marketplace listing gives 4.2 from 6,973 ratings.

Teamflect LTD is a United Kingdom company, incorporated on 27 May 2022. Teamflect gives 2018 as its founding year. Teamflect filed micro-company accounts made up to 30 April 2025. Bora Ünlü is the CEO.

Plan limits, unfinished features and AI duties that fall on you

Which AI features come with your plan

Teamflect’s AI page puts AI-powered features on every plan, including the free one. Its help centre limits some AI features to Professional. Fine-tuning the AI with company information is a Professional feature. So are the link for other AI assistants, development goals with AI, automatic AI translation and unlimited AI Agent use. Agent use is limited per user on Starter and Essential. Translation is in a controlled rollout. Ask Teamflect for a written list by plan.

AI meeting transcripts are marked “Coming soon”

Teamflect marks AI meeting transcripts “Coming soon” in its admin guide and in its user guide. Ask for a date before you count on this feature.

Your contract makes you responsible for checking the AI

Teamflect defines any use of its AI for employment or worker management decisions as “High-Risk Use” in its terms. You are solely responsible for validating AI outputs and decisions based on them. Qualified staff must review AI outputs before a decision with legal or similarly significant effects. You may not use AI as the only basis for a decision such as a performance rating. You must not use AI outputs to infer protected characteristics. You also agree to defend Teamflect against outside claims that come from relying on AI outputs without human review. AI components marked as Beta Services must not be used for decisions with legal or similarly significant effects.

The AI Act question is left to you

Annex III of the EU AI Act covers AI that monitors and evaluates people’s performance at work. Systems in Annex III are high-risk. An exemption under Article 6(3) can apply when the AI only does a preparatory task. It never applies to AI that profiles people. In its data agreement, Teamflect intends its AI features as general-purpose helpers. It does not intend them as high-risk systems under the EU AI Act. Under the same agreement, you are responsible for setting up the AI features to avoid high-risk use. You are also responsible for getting consent from the people concerned. You need that before you turn on AI features that process personal data.

The AI profile summary is hidden from the employee it covers

AI profile summaries give managers and admins an overview of a person’s role, reviews, goals, feedback and recognition. They are not visible to the employee they describe. Teamflect presents them as a way to identify high-potential employees. Whether the employee sees the AI-suggested nine-box placement depends on the template an admin chooses.

Two answers on where your data sits

Teamflect’s security page puts data on Microsoft Azure in five isolated regions. They are the United States, Canada, the Netherlands, Singapore and the United Arab Emirates. Its page written for AI chatbots puts all data inside the customer’s own Microsoft 365 and Azure accounts. Hosting in any Azure region that meets sovereignty rules is an Enterprise feature. Ask Teamflect which applies to you and where the AI processing happens.

Two uptime numbers

Under its service level agreement, Teamflect commits to commercially reasonable measures. The aim is at least 99% availability each month. Its security page gives a 99.9% uptime guarantee under the same agreement. Teamflect reports uptime and incidents on a public status page. Ask Teamflect which figure goes into your contract.

Training on your data has two exceptions

Teamflect rules out using customer inputs to train or fine-tune AI models. It does so in its AI policy FAQ and its data agreement. Under its master services agreement, Teamflect promises not to use customer data or prompts to train general AI models. It adds two exceptions. Ask Teamflect what they are. Teamflect may also keep a record of AI use. It may do this for security, to stop abuse, to fix faults and to check quality. Under its data agreement, Teamflect may monitor prompts and outputs to detect abuse, harmful use or breaches of law. Teamflect sets up its AI providers so they do not keep customer data or train on it.

Free for 10 users, then $7 to $11 per user per month billed annually

Starter is free for up to 10 users. Going above 10 means paying for all users, not only the extra ones. Essential costs $7 per user per month billed annually, or $9 billed monthly. Professional costs $11 billed annually, or $14 billed monthly. Enterprise is priced by quote. It adds an isolated copy of the service on dedicated cloud infrastructure. It also adds data migration from a previous solution. Retention and deletion rules can apply to each part, record or employee.

Here is a worked example. Take 200 licences billed annually. Essential costs 200 times $7 times 12, which is $16,800 a year. Professional costs $26,400. Professional billed monthly costs $33,600. Customers buy licences only for the people who will use Teamflect. Volume discounts and packages built around single parts of the product exist. Non-profits can get discounts of up to 60%. Teamflect charges no extra fees for implementation, support or training. It may update its pricing model or service tiers at its sole discretion with prior notice.

Subscriptions renew automatically for one-year terms. Either side must give 30 days’ written notice to stop that. Payment obligations cannot be cancelled and fees are not refundable. On annual plans, the licence count cannot be cut until the agreement ends. A downgrade takes effect at the next renewal. A monthly self-service plan needs 30 days’ notice before the next billing date. Ending the contract for a breach that is not fixed needs 30 days’ notice. If you end the contract that way, Teamflect refunds prepaid fees for the rest of the term. The laws of England and Wales apply. Liability is capped at what you paid in the twelve months before the first incident. Teamflect last updated the agreement on 2 December 2025.

On request, Teamflect makes your data available to export or download. That holds within the term and for 30 days after it ends. After those 30 days it has no duty to keep your data and deletes it. Admins can export user data and product data from the Admin Center. The files are Excel or CSV. Data export is an Essential feature. Review results can be downloaded in bulk as PDFs. Data entered during a free trial is lost for good unless you subscribe or export it. Under the licence model that covers everyone in the organisation, deleted users’ profiles stay searchable. Only global admins see the option to search them.

Teamflect’s page written for AI chatbots puts all parts in one price, with no features held back. Its help centre puts data export on Essential and the HR system link on Professional. Ask Teamflect for a written list of what each plan includes before you pick one.

You could also run review steps with agents built on your own systems. We have mapped what that looks like. The HR Agent Map

Twelve questions to put to Teamflect

Which AI features come with the plan we want, and how many AI Agent uses does each user get? Good answer: A written list by plan that matches both your AI page and your help centre.
What are the two exceptions to your promise not to train AI models on our data? Good answer: Both exceptions named in writing, with a way to opt out.
Does our data sit in our own Microsoft account or in your Azure regions? Where does the AI processing happen? Good answer: One answer, with the region named in the data agreement.
Does 99% or 99.9% apply to us, and what do we receive if you miss it? Good answer: One figure in the order form, with a written remedy.
Which fields does the HR system sync read, and can we leave employment and income data out? Good answer: A field list and a setting that excludes income data.
Do the AI profile summaries count as profiling under the EU AI Act? Profiling blocks the Article 6(3) exemption. Good answer: A written answer from your legal team that our own lawyers can read.
How can an employee see the AI profile summary written about them? Good answer: A set route that we can describe to staff.
Can we cut the licence count mid-term if headcount falls, and do we get any money back? Good answer: A written rule, and renewal at the licence count we choose.
When will AI meeting transcripts be available, and on which plan? Good answer: A date and a plan in writing.
How much notice do we get before you change the pricing model or the tiers? Can you cap the rise at renewal? Good answer: Notice well before the 30-day cancellation deadline, and a cap in writing.
Can we keep our data for longer than the 30 days after the contract ends? Good answer: A longer written export window, with every part of the product in Excel or CSV files.
Which customer count is current, more than 1,000 or more than 5,000? Which Microsoft AppSource rating is current, 4.2, 4.6 or 4.7? Good answer: One rating with its date, and a customer of our size on Professional that we can call.

Alternatives to Teamflect

Workleap, Lattice, Engagedly, Leapsome, Culture Amp and PerformYard also run review rounds.

Workleap’s AI assistant is reachable inside Slack or Microsoft Teams. Workleap starts at $4,999 plus $32 per manager per month, billed annually. Lattice sends notifications about reviews, 1-on-1 updates and surveys into Teams. Its Performance product costs $10 per seat per month, with a $4,000 minimum annual agreement. Lattice’s AI agent can also attend 1-on-1 meetings held in Zoom, Google Meet or Microsoft Teams.

Engagedly has a bot in Microsoft Teams, Outlook and Slack for feedback and goal updates. Its Performance Suite costs $5 to $8 per user per month billed annually, with a minimum of $7,500 a year. Leapsome sends review reminders in Microsoft Teams. Its Performance Reviews product costs €7 per employee per month, and the exact figure comes in the quote. The Leapsome platform is always part of the plan.

Culture Amp lets employees send and receive feedback in Teams or Slack. A customer can install the app for one of them, not both. Culture Amp Perform has its own Request a quote button on its pricing page. PerformYard costs $5 to $15 per person per month, billed annually. HR system connections and sign-in with a company login come with every PerformYard plan. It also connects to Slack or Teams.

If you buy nothing, your team carries on with the process it uses today. The free Starter plan lets a team of 10 or fewer test Teamflect at no cost first.

More in this guide

The Performance management software guide
Compare the 33 researched products how each of the 33 researched performance management products runs reviews, what its AI does, what it costs and what its contract says.
Mid-market companies which performance management products suit a company of 200 to 2,000 people, what they cost and what to ask before signing.
Enterprise companies which performance management products suit a company of 2,000 people or more, what they cost, how the contracts run and what to ask.
Frontline and deskless teams how performance reviews reach staff who do not work at a desk, what that costs and what to ask vendors.