TEMRIK / FUTURES LAB / 2026
The future is a working hypothesis.
82 specific predictions about AI, work, property and the built world. Each one comes with a signal to watch, a business move to consider, and a reason to change our mind.
THREE TO WATCH
- 11 / PROPERTY Listing briefs become structured
- 31 / OPERATIONS The morning brief replaces dashboard hopping
- 41 / CONSTRUCTION RFI drafts cite revision state
Published 19 October 2026 Next review: January 2027
THE METHOD
Useful forecasts should be falsifiable.
These are original TEMRIK working hypotheses, not promises, product commitments or investment advice. Time horizons are indicative. Confidence is qualitative. Some underlying tools exist today; the forecasts concern their adoption, reliability or fit in real workflows.
Use the filters to find an area you work in. Open a card to see what would make the idea worth pursuing, and what would make us reconsider.
THE OPPORTUNITY MAP
82 ways to look ahead.
Browse the field notes. Find a signal. Design a small test.
№ 01Personal AIA personal context ledger becomes usefulPeople will keep a portable record of preferences, commitments and decisions that assistants can inspect with permission.
Look for repeat use across at least two independent services.
Design exportable context with explicit scope and revocation.
If users repeatedly rebuild context rather than maintain a ledger, narrow the product.
№ 02Personal AIMemory gets an expiry dateAssistants will let people set retention dates for remembered facts instead of keeping indefinite profiles.
Watch for retention controls in mainstream assistants.
Make expiry and source visible on each memory.
If people cannot predict what is remembered after expiry, the control has failed.
№ 03Personal AIHousehold agents coordinate calendarsShared household assistants will propose schedules across family members while preserving individual permission boundaries.
Watch multi-person conflict resolution in ordinary calendar products.
Prototype consent-aware scheduling, with each person approving their changes.
If consent overhead exceeds saved coordination time, do not automate.
№ 04Personal AIPersonal AI produces a decision receiptPeople will expect a short explanation of the evidence behind important assistant suggestions.
Watch user research on whether receipts change decisions.
Use source and uncertainty panels as default for consequential recommendations.
If receipts are ignored or misleading in tests, simplify the evidence view.
№ 05Personal AIAssistants prepare service enquiriesA person will ask an assistant to assemble a concise brief before contacting a builder, broker or adviser.
Watch enquiry forms accepting structured, user-approved summaries.
Let TEMRIK accept a brief with provenance and missing fields.
If suppliers still have to re-ask every question, the format has not worked.
№ 06Personal AIVoice becomes a review surfaceVoice assistants will read back proposed actions and ask for confirmation during hands-busy work.
Watch reliable interruption and correction in noisy environments.
Build spoken summaries for low-risk review, with a visual fallback.
If recognition errors survive confirmation, keep the task text-only.
№ 07Personal AIPeople carry permission profilesUsers will reuse a personal set of default AI permissions across apps, then adjust per task.
Watch interoperable permission export standards and user demand.
Keep permissions portable and understandable at the action level.
If portability adds ambiguity, local permissions remain safer.
№ 08Personal AIAI helps people compare advicePersonal assistants will show where two professional recommendations differ and which assumptions explain the gap.
Watch tools that preserve original adviser language and citations.
Offer comparison views without replacing professional judgment.
If summaries erase material caveats, stop the comparison workflow.
№ 09Personal AIAgent handoffs become visibleUsers will see when one assistant delegates to another tool or model.
Watch consumer products expose handoff logs.
Show the actor, permission and result for each handoff.
If the handoff log cannot be understood in usability tests, reduce delegation.
№ 10Personal AIPrivate context vaults become a purchase criterionSome buyers will choose assistants based on how clearly they can inspect and remove personal context.
Watch procurement questions and consumer churn tied to data control.
Treat context access and deletion as product features.
If buying decisions show no preference, keep investment proportional.
№ 11Agency & propertyListing briefs become structuredAgencies will collect property facts, source documents and uncertainty before generating campaign copy.
Watch documented review workflows at agencies.
Offer an evidence-backed listing intake template.
If staff find the brief slower than current practice, reduce required fields.
№ 12Agency & propertyCampaign variants require claim reviewAI will produce many ad variants, but agencies will gate factual property claims before publishing.
Watch approval logs and corrected listing claims.
Build claim-level review with source links and approver identity.
If reviewers cannot spot unsupported claims faster, the gate needs redesign.
READING DESK
The evidence beneath the questions.
These resources inform the context and method; they do not validate each forecast. The hypotheses and implications above are TEMRIK’s own.
- Stanford HAI · 2026 AI Index
Current research on AI capabilities and adoption.
- NIST · Generative AI Risk Management Profile (2024)
A practical frame for testing risk and controls.
- buildingSMART · IFC 4.3 documentation
An existing interoperability reference for built assets.
- Lenny’s Podcast · Josh Woodward (2026)
Inspiration for a lab mindset, not the source of these predictions.
FROM WHAT IF TO WHAT NEXT
Choose one forecast.
Test it in your world.
TEMRIK helps teams map the workflow, evidence and human authority needed to explore AI safely in real work.
Work with TEMRIK