AI Agents
Will AI Agents Replace Traditional Software?
Don't just read what happened. See what could happen next.
Prediction
AI agents will displace a meaningful share of task-centric SaaS workflows within 12–24 months, while core systems of record remain human-owned.
- Confidence
- 71%
- Horizon
- 12–24 months
- Impact
- High
- Direction
- Increasing
Prediction changed +20 points in 21 days
Short answer
AI agents are increasingly capable of performing multi-step work that once required dedicated apps — drafting, researching, filing tickets, updating CRMs. Replacement will be uneven: thin workflow tools are most exposed; deep systems of record and regulated stacks will absorb agents as interfaces rather than disappear.
Why this question matters
If agents become the default interface to work, the software value chain shifts from seats and dashboards to outcomes and orchestration. That affects enterprise IT budgets, startup defensibility, and which platforms control the agent runtime.
What's happening now?
Agent productization accelerates
Major labs and suites ship agent modes that chain tools, browse, and write back into business systems.
Signal · strong
SaaS vendors embed copilots first
Incumbents add agent features to defend seats rather than ceding the layer to a new interface company.
Signal · strong
Reliability and audit gaps persist
Enterprises still gate autonomy behind human approval for irreversible or regulated actions.
Signal · moderate
Pricing experiments move beyond seats
Outcome- and usage-based packaging appears beside classic per-seat SaaS.
Signal · moderate
What could happen next?
Scenario A
Accelerating adoption
Agents become the primary work surface for knowledge tasks; many point solutions consolidate into agent + platform stacks.
Scenario B
Gradual co-existence
Agents sit on top of existing SaaS. Replacement is limited to narrow workflows; vendors survive by owning data and compliance.
Scenario C
Adoption stalls
Error rates, security incidents, or regulation slow autonomy. Agents remain assistive copilots for years.
Key companies / entities
- OpenAI
- Anthropic
- Microsoft
- Salesforce
- ServiceNow
Evidence
- company
Enterprise agent rollouts inside productivity suites
Suite vendors shipping agent orchestration and connectors.
- research
Research on tool-using LLM reliability
Benchmarks still show brittle multi-step failure modes.
- news
Procurement interest in outcome-based AI pricing
Prediction history
- Sep 22, 202671%
- Sep 15, 202664%
- Sep 8, 202657%
- Sep 1, 202651%