Why the Next Generation of Enterprise Software Looks Nothing Like Salesforce

Technological/Product Concepts and “Enterprise Software Beyond Salesforce”

Lightfield’s positioning

Lightfield is positioned as an enterprise CRM for AI work, but with a broader goal than a traditional “system of record.” The core idea is to model “customer reality” so AI agents can use it to take action—such as tool calls, prompts, and automation.

Business World Model (key technology concept)

• Converts customer emails, calls, and meetings into a structured record that AI agents can operate on.
• Core claim: a CRM shouldn’t only store data; it should reorganize messy reality into something both machines (and humans) can understand.

Canonical “activity log” as the primitive

• Lightfield builds a chronological activity log capturing relationship events such as:
  • outreach, messages, meetings
  • documents
  • product usage signals
  • payments
• This log becomes the canonical foundation used to trigger traditional CRM updates like fields/stages.

Semi-structured + unstructured storage tradeoff

• Fully unstructured storage was too slow for querying (the “needle in a haystack” problem).
• Instead, Lightfield uses semi-structured storage:
  • keeps large unstructured material in the activity log
  • enables inference and downstream querying

Schema-less setup for onboarding

The initial configuration is designed to be “schemaless”:

• Connect email, call recorder, and data warehouse sources.
• Lightfield assembles relationships and enriches in real time.
• If fields/stages change later, the system can re-traverse the activity log to refill derived structure.

Message: “intelligence is greater than schema.”

Inference-driven CRM functions (example workflow)

For Customer Success questions like “Is this account ready for expansion?”, Lightfield can:

• Review interactions (what they said, tickets, etc.)
• Check product usage (e.g., infer last login time from activity-log entries)
• Traverse CRM entities/schema, then dive into per-customer logs to recommend:
  • expansion timing
  • which products to expand with

Product Features and AI/Automation Implications

AI agent workflows instead of rigid UI-only processes

Go-to-market “sequence” setup shifts from manual rule building (arrows/conditions/variables) to conversational agent-driven “recipes” that adapt to the world model. The agent can generate and run workflows using context from modeled business reality.

Multiple interaction modes, not one “religious” UI

Lightfield supports dashboards and tables, but it also aims to enable more agent/AI-driven execution (including a possible CLI-style mode).

Support for arbitrary/custom data models

• Began with fully arbitrary custom objects/relationships to serve companies with unusual business models.
• Emphasized the architecture doesn’t force a single CRM ontology.

Adoption and GTM (Brownfield vs Greenfield)

Wedge into enterprise CRM (brownfield vs greenfield)

Lightfield argues the wedge isn’t replacing Salesforce at the “send emails” level. Instead, it wins by better modeling of the company + customer reality, enabling smarter decisions and automation.

Negative pricing tactic (early customer acquisition)

Since it was difficult to get adoption of a new CRM directly, Lightfield used negative pricing:

• Offered space for startups to use the product (plus feedback).
• Used continuous feedback to iterate and improve.

Adoption hurdle: “VC/product believers” vs “Sales VP training”

Even when leaders like the AI system, a new VP of Sales may resist due to Salesforce training and habit. Lightfield’s strategy:

• Give broader access for free to the whole company.
• Let non-rep teams (engineering, finance, CS, etc.) generate internal network effects—making switching harder.

Pricing Analysis (Clear Concrete Takeaways)

Pricing experimented across extremes

• Started with seat pricing, then found consumption skewed massively (tail vs head usage).
• Tried pure consumption/credit pricing, which led to inactivity (users signed up but didn’t engage).

Final model described

• Platform fee + seats for core CRM work, such as:
  • capturing meetings
  • filling records
  • updating tasks
• Consumption pricing for other areas, especially:
  • pipeline generation (enrichment / “alpha” that improves meetings)
  • workflow automations (real work like routing lead/demo requests)
  • intelligence/forecasting (scenario planning; “pay for the alpha”)

Outcome-based pricing skepticism

Lightfield notes outcomes are hard to tie directly to software because results depend heavily on:

• product-market fit
• broader market conditions

Conclusion: charge for work, not guaranteed outcomes.

Company Culture and Velocity (Process and Tooling Concepts)

Pivot speed + fewer rigid “swim lanes”

A prior venture suffered from rigid ownership across product/marketing/CS, slowing pivoting. Lightfield avoids this by:

• having no swim lanes
• enabling everyone to participate in product/customer-success problems

Operating model

• Everyone attends the same daily standup.
• Stack-rank the most important problems; whoever is free picks them up.
• Continuous planning:
  • daily list changes
  • weekly re-prioritization

High leverage due to AI tooling

AI tooling helps teams ramp faster because the LLM/automation can interact with tools. Examples mentioned include:

• Figma via LLM
• Linear for task creation

Risk/worry: speed as a competitive constraint

They worry competitors could win by shipping dashboards/workflows faster. They reference a cautionary story about startup delays (CRM/dashboard delay → moved to Salesforce).

Example: “What Can Lightfield Do Today?” (Most Concrete)

Power (health/pharma trial matching)

• Models both B2B and B2C sides in Lightfield.
• Scrapes FDA and ClinicalTrials.gov to build a world model of trials.
• Matches trials with participants and pharmaceutical partners.
• Reported outcome: helped an Alzheimer’s seeker find a frontier treatment within days.

Expansion readiness signals for Customer Success

Uses:

• interaction logs
• product usage to answer expansion timing and prioritization questions.

Main Speakers / Sources

• Keith Paris — CEO at Lightfield (primary source)
• Joe Schmidt — host (A16Z podcast)
• Alex Rumpel — host/partner (A16Z podcast)

Mentioned reference:

• Andreessen Horowitz (A16Z) podcast
• Lightfield’s founding team background (not a separate speaker)