Team decisions / Changing delivery
Let the evidence change the engagement.
AI tools, specialist capability and changing buyer expectations are reshaping sourcing discussions. The useful question is what should change in your own arrangement. Separate a market signal from a supplier claim, then inspect the work, costs and responsibilities before changing the commitment.
A decision, not a prediction
Changing tools do not decide who owns the work.
The future of software outsourcing is best assessed through changes in how work is produced, checked and carried forward. Current research gives reasons to examine AI use and sourcing expectations. It does not decide your team size, contract, budget or delivery model. Use those signals to ask better questions, then make changes on evidence from the actual engagement.
- Observed signal
- Name the source, date, population and what was measured. A stated spending intention, a reported benefit and an observed delivery result answer different questions.
- Conditional implication
- Explain what would have to be true for the signal to matter here. Faster drafting may change the work mix only if review, integration and acceptance can handle it.
- Buyer decision
- Specify the evidence needed to change an assignment, scope or commercial commitment. Keep an owner and a review point for the assumption.
Five areas to reassess
Ask what changes in the delivery arrangement.
These are decision areas informed by current research and established sourcing guidance. Their relevance depends on the product, organization and supplier. They are not five guaranteed market outcomes.
AI-assisted production and review
01Changing question: Which part of delivery improves when the team uses AI?
- Evidence basis
- Google's September 2025 DORA summary describes broad AI use among nearly 5,000 surveyed technology professionals and reported individual benefits, while noting organizational complexity.
- Local question
- Where does time go between a requested change and accepted behavior?
- Practical response
- Trace a representative change through drafting, review, correction, integration and acceptance.
- Retained decision
- The engineering owner sets the acceptance standard and decides whether the result supports a change.
- Inspect
- Comparable work scope, review effort, rework and accepted behavior.
- Unsupported leap
- Tool adoption is treated as proof of fewer people, lower prices or reliable output.
- Revisit when
- The tool, task mix or verification burden changes.
Skills around the generated work
02Changing question: Can the assigned people explain and verify what they deliver?
- Evidence basis
- DORA's AI Capabilities Model author summary connects AI outcomes to technical and organizational conditions, including internal context and clear expectations.
- Local question
- Who understands the business rule, architecture and failure consequence?
- Practical response
- Ask contributors to explain a decision, diagnose a failure and repair a change in the relevant system.
- Retained decision
- Client technical and product owners decide which expertise the assignment needs.
- Inspect
- Reasoning tied to the code, relevant checks and a maintainable correction.
- Unsupported leap
- A tool badge or a generated demonstration is treated as evidence of independent engineering judgment.
- Revisit when
- The work crosses a new domain, integration or operating boundary.
Different boundaries for different work
03Changing question: Which responsibilities should be internal, external or shared?
- Evidence basis
- The GOV.UK Sourcing Playbook evaluates internal, external and mixed delivery options, including capability, transition and whole-life cost.
- Local question
- Which part of the service can another party own and which decisions need to stay close to the business?
- Practical response
- Compare arrangements for the same service boundary and receiving responsibilities.
- Retained decision
- The buyer owns the delivery-model decision and its unresolved dependencies.
- Inspect
- A responsibility map and equivalent cost and transition assumptions.
- Unsupported leap
- A fashionable engagement label is assumed to fit every part of the product.
- Revisit when
- The scope, internal capability or supplier responsibility changes.
Knowledge that remains usable
04Changing question: What will the buyer still understand after the assignment changes?
- Evidence basis
- Whitelane's June 2026 UK and Ireland summary identifies retaining critical knowledge as a reason respondents give for reducing external reliance.
- Local question
- Who can explain an important rule and continue the work without the original contributor?
- Practical response
- Keep a receiving owner involved and test one realistic maintenance task.
- Retained decision
- The client technical owner accepts knowledge continuity; qualified owners resolve rights and access.
- Inspect
- Current decisions, usable instructions and a receiving-person result.
- Unsupported leap
- An internal repository is assumed to establish practical independence.
- Revisit when
- A contributor, platform or delivery boundary changes.
Expectations beyond staffing volume
05Changing question: Can the supplier challenge the proposed work with useful evidence?
- Evidence basis
- Whitelane's same regional survey summary reports buyer concerns about insufficient challenge, business knowledge and value from providers.
- Local question
- Which assumption should the supplier question before implementing the request?
- Practical response
- Ask for an alternative, its tradeoffs and a way to judge whether the change helps users.
- Retained decision
- The business owner chooses the intended result and accepts the tradeoff.
- Inspect
- A decision record connecting the work to a user or operating need.
- Unsupported leap
- An outcome-based label is mistaken for an agreed measure or a result the supplier can control.
- Revisit when
- The business priority, acceptance evidence or dependencies change.
Read the evidence at its own scale
Keep a signal separate from a commitment.
Whitelane's 2026 UK and Ireland study covers nearly 400 large IT-spending organizations. Its summary reports that 32% expect higher external IT spending, 30% no change, 18% a reduction and 20% are uncertain. These are regional intentions across IT sourcing, not realized software outsourcing growth. Use the distinction below for other claims too.
| Evidence type | What it can show | What it cannot settle | Buyer question | Next evidence |
|---|---|---|---|---|
| Sourcing survey | Stated plans in a defined population | Your future spend or supplier capacity | Does this population resemble our situation? | Current scope and market conversations |
| AI research summary | Reported use, experience and associations | A causal saving for this team | Which delivery stage could improve? | Comparable local work and acceptance records |
| Capability guidance | Practices worth examining | A guaranteed implementation result | Which condition is missing here? | A bounded change and observed result |
| Supplier proposal | An offered scope and commercial basis | Proven delivery performance | Which assumptions drive this promise? | Relevant work evidence and exclusions |
| Local trial | Behavior under the tested conditions | Every future task or operating condition | What remains untested? | Repeat affected checks before expanding |
| Renewal decision | An authorized commitment and its basis | Permanent suitability | What would make us revisit it? | Named review trigger and current evidence |
Work through a renewal proposal
Test the claimed improvement before changing the commitment.
Illustrative situation: a supplier proposes a different staffing and price arrangement because it now uses AI-assisted development. Choose a bounded change, such as adding an approved account-status rule to an existing workflow. The exercise informs a decision; it is not a controlled experiment or a promised saving.
- 01
Write the claim precisely
Ask which task changes, why the proposed assignment can cover it and which assumptions affect the price. Record exclusions and unresolved conditions.
- 02
Set the comparison
Use comparable previous work where it exists. Agree the rule, failure cases and acceptance evidence. Record differences that make a direct comparison unreliable.
- 03
Observe the full change
Record implementation, review, correction and integration effort. Include buyer time and tool costs. Have the authorized owner inspect the actual behavior.
- 04
Check the receiving capability
Ask another authorized person to explain the rule and make a small permitted adjustment. Locate the relevant decisions, dependencies and checks.
- 05
Revise only what the evidence supports
Agree any change to scope, assignment or commercial terms through the authorized process. Keep limitations visible and set a review trigger for a different task mix or rising rework.
Prepare without guessing the market
Make the next decision easier to revise.
A sourcing arrangement needs room to respond to new evidence. These are practical buyer actions, not claims that every provider already works this way.
- Keep a delivery baseline
- Retain the scope, checks, accepted behavior and known limitations for meaningful work. Changing tool use is easier to assess when the previous arrangement is understandable.
- Agree the permitted tool boundary
- Name which tools and data uses are authorized, who reviews changes and where decisions are recorded. Include supplier contributors in the same practical understanding.
- Develop receiving expertise
- Give internal owners time to learn the system and participate in review. Buying a service still requires someone able to judge its scope and results.
- Record the reason to reconsider
- Specify a trigger such as new sensitive data, repeated failed checks, a new dependency or a changed business objective. Review the affected arrangement when it occurs.
Questions about the next engagement
Ask for the decision behind the forecast.
A useful answer identifies what is known, what is conditional and what you can check locally.
- Will AI replace software outsourcing?
- The sources here do not establish that outcome. AI can change how parts of the work are performed, while businesses still need decisions, integration, verification and operation. Examine which responsibilities change in your engagement rather than treating a forecast as a staffing plan.
- Should AI use automatically reduce the supplier's price?
- No automatic reduction follows from using a tool. Inspect the offered scope, review effort, rework, tool costs and retained responsibilities. Any change to commercial terms needs a clear basis and the authorized agreement process.
- Will every contract become outcome-based?
- There is no basis here for that prediction. A useful outcome needs a definition, a measure, dependencies and an owner able to influence it. Assess the actual arrangement and how payment relates to responsibility.
- Which capabilities should the buyer keep?
- Keep enough business and technical understanding to set priorities, assess evidence and continue making informed decisions. The exact roles depend on the product and delivery boundary. Test whether another authorized person can understand and maintain important work.
- What should we change before the next renewal?
- Make the present scope, acceptance evidence, costs and ownership understandable. Ask the supplier to identify a specific improvement and how to test it. Use the result to revise the affected commitment, with explicit limits and a reason to review it again.
Source basis
Research signals and guidance behind the questions.
- 01
Google
Inside the 2025 DORA reportSeptember 2025 author summary of a global technology-professional survey. Reported experiences and associations inform the AI discussion; they do not establish causal savings for an outsourced team. Full report not used as reviewed evidence.
- 02
Google Cloud DORA researchers
Introducing the DORA AI Capabilities ModelSeptember 2025 author explanation of the model and its interview and survey basis. Informs questions about organizational conditions for AI-assisted work, not guaranteed performance.
- 03
Whitelane Research
2026 IT Sourcing Study: UK and IrelandJune 2026 public summary. Large IT-spending organization sample, regional scope and stated intentions are retained. Provider rankings and forecasts are not generalized to this buyer or to software outsourcing worldwide.
- 04
GOV.UK Cabinet Office
The Sourcing PlaybookDelivery-model assessment and whole-life cost sections reviewed. Government guidance supplies evaluation questions about capability, assets and transition; public procurement rules and older reform passages are not adopted as universal requirements.
Start with one real workflow
A Systems Audit is the usual starting point. If the opportunity is already clear, we can move directly into a focused build.
Show Us the WorkflowStart with the free automation readiness checklistOBSERVEQUANTIFYDECIDEBUILD