A colleague used to know which policy applied, where the right file lived and who could answer a difficult internal question. TD Bank is looking for ways to put more of that knowledge into AI tools.
On September 22, TD announced a collaboration with Cohere worth up to C$25 million over three years. The announcement is about building and testing useful applications. For employees, the pressure starts when a task that once justified hours of paid work takes minutes.
Cohere and Layer 6 are starting with bank knowledge
TD says Cohere’s enterprise AI specialists will work alongside its Layer 6 team at Toronto’s MaRS Discovery District. They will explore applications across the bank, with knowledge management named as one area of focus. TD says it wants greater productivity and stronger decisions.
In an employee’s day, knowledge management can mean locating a current policy, pulling facts from files or finding an answer another team has already written. A reliable tool could shorten that search and draft a response for a person to check. Those paid minutes are the first thing management can squeeze. TD has not published the exact Cohere applications, the teams that will use them or a rollout date.
TD has already put a price on faster work
At Investor Day in September 2025, CEO Raymond Chun set an ambition of C$1 billion in annual AI value: half from revenue and half from cost savings. He said TD was reshaping its top 20 processes, which represent about 60% of processing costs, using AI and automation to deliver the savings. TD later reported about C$170 million in AI value from roughly 75 use cases in fiscal 2025.
The Cohere collaboration is a new investment within that wider AI strategy. TD has not assigned any portion of the C$500 million target to Cohere. Management can use saved hours to handle more work, raise output targets or lower costs. Employees need to know which result their team will be measured against.
For TD’s broader staffing record and the CIBC comparison, read our TD and CIBC investigation.
The first work exposed is preparation around a decision
The near-term pressure falls on repeatable preparation. These are the tasks an internal knowledge tool could speed up before anyone changes an entire job title. The rows below are our assessment of task exposure, not TD’s list of roles selected for reduction.
| Work people do | What may get faster | What an employee should ask |
|---|---|---|
| Policy and procedure support | Finding the right version of a rule and answering repeat internal questions | Who checks the answer when the rule has an exception? |
| Operations and document support | Reading files, extracting facts and drafting case summaries | Will case volume or turnaround targets rise? |
| Lending preparation | Assembling facts for an underwriter to review | Which checks still require a person, and who owns errors? |
| Internal reporting | Collecting standard facts and drafting routine updates | How will the time saved be reflected in team workload? |
Judgment still has an owner. A wrong policy answer, an unusual borrower file or a missed control can cost the bank more than the minutes saved. Employees whose work includes handling those exceptions need to make that responsibility visible.
TD’s mortgage tool shows what a shorter task looks like
TD has already reported that a separate Layer 6 mortgage tool cut the average turnaround for a pre-adjudication step from 15 hours to under three minutes. It classifies documents, extracts information, checks selected policy requirements and prepares a summary for an underwriter. Those are elapsed process times, not paid hours removed or a result of the Cohere deal. They show the kind of preparation TD knows how to compress.
WorkiQ measures time on another track
Reuters reported in June that TD planned to use WorkiQ with some financial crime and risk staff to measure time in browsers, chat and meetings. TD said the software helps managers see workflows, capacity and performance. WorkiQ is separate from Cohere and TD said it is not AI. Together, the public disclosures show a bank paying close attention to both the work and the time it takes; they do not establish a shared system or a layoff plan.
Three changes TD employees should watch
First, watch what staff are asked to put into a shared knowledge base. A pilot that absorbs years of policy answers and case notes tells you which expertise management wants the tool to retrieve.
Next, watch the measures used after a pilot: cases per employee, time per file, response speed and the number of routine questions sent to a specialist. A faster answer has a workforce consequence when targets change.
Then look at who handles exceptions. If the team is left with harder files while expected to complete more of them, ask what training, review time and staffing support come with the new target.
Make the hidden work visible before targets move
Keep a private record of the decisions only you can make, the policy conflicts you resolve and the errors you catch. Record how long those cases take without taking confidential client material home. That gives you a concrete account of your work when a dashboard reports only faster average handling time.
Ask your manager which tasks the pilot is meant to shorten, how quality will be checked and whether the team’s output target will change. Refresh your résumé while you have time to choose your next move.
Three free tools for TD employees
Take the free two-minute Job Threat Check to assess what is changing around your role and manager.
Get the free Weekly Layoff Intelligence Report for verified bank workforce signals and what to watch next.
Follow TD and other major employers with the free Layoff Tracker + Corporate Stress Index, which tracks public evidence of workforce pressure.
Sources and evidence
Sources reviewed September 24, 2026. The Cohere terms come from TD’s announcement. The C$500 million target comes from Chun’s earlier Investor Day remarks. The mortgage example is a separate TD tool. The WorkiQ account comes from Reuters. Task exposure and employee warning signs are our analysis.
- TD: Cohere and Layer 6 collaboration, September 22, 2026 — Investment of up to C$25 million, knowledge management focus and co-located experts.
- TD: Raymond Chun’s 2025 Investor Day transcript — Annual AI value ambition, cost-savings half and top 20 processes.
- TD: Fiscal 2025 fourth-quarter earnings transcript — TD’s reported C$170 million in fiscal 2025 AI value from roughly 75 use cases.
- TD: Layer 6 mortgage pre-adjudication announcement — A separate TD tool and its measured process turnaround.
- Reuters: TD’s WorkiQ monitoring plan — June reporting on a separate non-AI tool for some risk and financial crime staff.
About The Grind Hotline
The Grind Hotline is a worker-first global media and workplace intelligence platform and business podcast covering layoffs, AI job pressure, restructuring and decisions affecting job security. It reaches people in more than 100 countries.
Harj Singh, The Host, is an ex-banker and former Fortune 100 and Fortune 500 sales leader. He lost his job twice in five years and built The Grind Hotline to help employees see pressure earlier.
The platform won a 2026 dotCOMM Platinum Award for Content Strategy and a 2026 MUSE Creative Awards Silver award in Branded Content, Cause/Awareness. Read its Media and Editorial Standards.
Important Disclaimer
This article offers general workplace information from public documents and credited reporting. It cannot predict an individual employment outcome or replace legal, financial or career advice.