The soil microbiome has become a product category: seed-applied bacteria, fungi and engineered microbes sold as inputs that promise higher yield or less synthetic fertilizer and fungicide. In September 2026, three companies announced a launch or financing, and an established fourth still leans on its own claims; every performance number cited here came from the company itself. This article lays out what each product claims, how it is meant to work, and which pieces of evidence a skeptical reader would still want.

What a microbial crop product actually is

Three families sit under the same heading, and mixing them up causes most of the confusion.

  • Inoculants. Beneficial bacteria, usually rhizobia for legumes, applied so that nodules form on the roots and fix atmospheric nitrogen.
  • Biostimulants and seed-applied biologicals. Mixtures of bacteria, fungi or signaling chemistry meant to help roots take up nutrients or tolerate stress.
  • Engineered or biocontrol microbes. Organisms modified, or computationally designed, to fix nitrogen or to suppress a disease.

The shared difficulty is that soil is crowded. A strain added at planting must compete with the organisms already there, and the outcome varies by field, year and weather. That is the background for each claim below.

Sound Agriculture’s INFLECT: a seed biological for high-ammonium corn

Per AgFunderNews, Sound Agriculture released INFLECT in September 2026. It is aimed at corn grown with anhydrous ammonia or heavy manure, conditions the company says can reduce soil microbial diversity. It is a powder added in the planter box. The company describes two Bacillus strains meant to buffer pH near roots and promote root growth, plus a fungus, Coniochaeta nivea, said to break down manure and complex compounds to release nutrients.

The reported trial covered about one year and 12 sites in Illinois, Kansas and Missouri. Yield rose by about 6 bushels per acre on average and improved at 11 of the 12 sites. The article gives no total acreage, no trial design, no controls, no statistical analysis and no price. All figures came from the company’s agronomy lead. Eleven of twelve sites sounds persuasive, but with undisclosed methods and a single season it is a promising signal and not a result.

Lilac Agriculture: why inoculants sometimes fail

Lilac’s story explains the field problem well. In a September 22, 2026 report, the company said native soil bacteria often outcompete the strain a farmer applies, and that many commercial rhizobia date back to older USDA collections. A co-founder reportedly found that the bacteria inside nodules were not the strain in the product the farmer had bought.

The company screens strains from North Dakota State University’s collection for nitrogen fixation and competitiveness, then uses molecular markers to confirm which strains occupy nodules. Its first strain, PL11, is naturally selected, not gene-edited. It claims an 11% aggregate yield increase over no inoculant and 6% over an existing commercial one, sold on a small number of pea and lentil acres. The company raised an oversubscribed $2.3 million seed round.

The nodule-verification step is the interesting design choice. A product that can show it colonized the root is easier to evaluate than one that only reports end-of-season yield. The yield percentages, again, are company-reported, and the article does not describe trial design.

Pivot Bio: the engineered-nitrogen benchmark

Pivot Bio is the best known engineered-microbe nitrogen company. Its website says its gene-edited microbes convert nitrogen from the atmosphere into ammonia on the plant’s root system and are meant to complement, not replace, a fertilizer plan. It lists products for corn, wheat and small grains, cotton and sorghum. The company says it has avoided 1,050,000 metric tons of CO2e since 2022.

The homepage offers no supporting data for its performance statements and does not state acreage. An AgFunderNews report from July 2021 said Proven was designed to let corn growers cut 40 pounds of synthetic nitrogen per acre and that Pivot had raised $430 million in a Series D, with more than a million acres claimed. Those are five-year-old company statements, so I use them only for background. I did not find a 2026 independent field analysis.

Robigo Bio: designed microbes against disease

Robigo Bio’s Series A (September 17, 2026) shows a different bet. It starts with a microbe chosen for shelf stability, then computationally adds traits so that, applied once as a seed treatment, the organism makes and delivers an active ingredient through the season. Two products are near launch: a soybean treatment for Sudden Death Syndrome, slated for 2028, and one for fusarium wilt in lettuce and berries, slated for 2029.

The company reports that its soybean product is on track to match commercial chemistries at half the cost and that the fusarium product showed a threefold yield increase over untreated controls. Its first field trials in 2025 were cleared by the USDA and EPA, and it has started registration for the soybean product. These are early company numbers, and the launch dates are years out.

The nitrogen idea in plain terms

Most of these products circle one goal, getting plants more nitrogen without more fertilizer. Air is about four-fifths nitrogen gas, which plants cannot use directly. Certain bacteria carry an enzyme that converts it to ammonia, a form plants can absorb. Legumes host such bacteria in root nodules, which is the textbook partnership that inoculants try to improve. Cereal crops such as corn have no comparable nodules, so a company that wants to supply nitrogen to corn has to place bacteria on or near the roots and hope they release enough ammonia to matter.

That is what Pivot Bio says its gene-edited microbes do, by its own account. The difficulty is a trade-off: a microbe that releases a lot of nitrogen spends energy doing so, and it still has to survive among thousands of competing species. Whether any particular product shifts the nitrogen budget of a whole field is an empirical question, answered only by well-designed trials.

Claims at a glance

Company Approach Reported result Evidence gap
Sound Agriculture Bacillus strains and a fungus, seed-applied About 6 bu/acre average gain, 11 of 12 sites No design, controls or statistics disclosed
Lilac Agriculture Selected rhizobia, nodule verification 11% over no inoculant, 6% over a commercial one Trial design not described
Pivot Bio Gene-edited nitrogen-fixing microbes Company states emissions avoided No supporting data on its homepage
Robigo Bio Computationally engineered disease-control microbes 3x yield on fusarium product vs untreated Early; launch 2028 to 2029

How to judge a microbial input

  1. Ask who ran the trial. Company-run trials are normal at this stage, but independent university trials weigh more.
  2. Look for replication. One season and 12 sites is a start. Multiple years and varied soils tell you about consistency.
  3. Check the comparator. A gain over no treatment is easier than a gain over a commercial standard.
  4. Look for proof of colonization. Lilac’s nodule markers are one way to show the microbe did its job.
  5. Read the regulatory status. A field-trial clearance is not a registration.

None of this means these products fail. It means the public record is still mostly press releases and trade interviews. A grower or agronomist deciding on a specific product should consult local extension trials and an agronomist; this article is general information, not agronomic or investment advice.

Why good evidence is hard to produce

Field biology is noisy. Two fields a mile apart can differ in soil texture, pH, residue, moisture and the resident microbes, and a season’s rain can swamp a modest treatment effect. A fair test needs replicated strips, untreated controls beside treated ones in the same field, and several seasons. That costs money and time, which is why early-stage companies usually publish results from their own demonstrations.

A gain of 6 bushels per acre, as reported for INFLECT, is a small percentage of a typical corn yield, so it can disappear into field-to-field variation unless the trial is designed to detect it. The article does not give the variability, so a reader cannot tell whether 6 bushels is a clear signal or a lucky average. The same logic applies to percentage gains for legumes.

Colonization is the second hard part. Lilac’s complaint that the strain in the nodule differed from the strain in the bag is the kind of check that growers rarely see. If a product cannot show the organism established itself, any yield change is hard to attribute to it.

The wider shift

AgFunderNews placed these companies inside a move away from chemical inputs. Its May 2026 roundup lists, among others, Kula Bio, which works with a modified Xanthobacter autotrophicus microbe to fix nitrogen as a fertilizer replacement, and Resurrect Bio, which restores plants’ own immune receptors through gene edits. It also warns that most figures in the piece were company claims and that several featured firms are AgFunder investments. That disclosure should travel with any reading of this coverage: trade outlets with investor ties are useful for finding out what exists, but they are not independent verification of whether it works.

For a farmer, the practical step is to ask a supplier for third-party data and for local extension trial results. For a general reader, the useful takeaway is simpler: the category is real and funded, the mechanisms are plausible, and the proof is still arriving.

Sources